A quantification model of Traditional Chinese Medicine syndromes in children with idiopathic precocious puberty and early puberty

Online Submissions:http://www.journaltcm.com J Tradit Chin Med October 5; (5): 6-66 info@journaltcm.com ISSN 55-9 JTCM. All rights reserved. CLINICAL STUDY TOPIC A quantification model of Traditional Chinese Medicine syndromes in children with idiopathic precocious puberty and early puberty Yanyan Lin, Yuanyuan Xue, Jian Yu, Yonghong Wang, Wen Sun, Yanyan Sun, Yumin Shi, Yiqun Zhang, Hong Hu aa Yanyan Lin, Yuanyuan Xue, Jian Yu, Yonghong Wang, Wen Sun, Yanyan Sun, Yumin Shi, Yiqun Zhang, Hong Hu, Department of Traditional Chinese Medicine, Children's Hospital of Fudan University, Shanghai, China Supported by the National Natural Science Foundation of China (No. 878); the Shanghai Science and Technology Research Grant Program (No. 9dZ976); State Key Clinical Department of TCM pediatrics Correspondence to: Prof. Jian Yu, Department of Traditional Chinese Medicine, Children's Hospital of Fudan University, Shanghai, China. yuj@shmu.edu.cn Telephone: +86--699; +86-8755699 Accepted: April 8, Abstract OBJECTIVE: To establish a quantification model of Traditional Chinese Medicine (TCM) syndromes by sampling patients undergoing idiopathic precocious puberty (IPP) and early puberty. METHODS: A questionnaire for classifying and quantifying TCM syndromes was designed and administered. All the results were analyzed; the relationship between types of syndrome and 7 symptoms were summated. Meanwhile, the frequency distribution of each symptom or sign was aggregated. Fuzzy mathematics was used to develop a quantification model of TCM syndromes. RESULTS: We found that precocious puberty had types of syndrome, including hyperactivity of fire due to Yin deficiency (Syndrome I), depressed liver Qi transforming into fire (Syndrome II), and end retention of damp heat (Syndrome III). In the IPP group, Syndrome I was the most common principal syndrome (%). Forty-six patients (.8%) were diagnosed with Syndrome I accompanied by Syndrome II and (.8% ) were diagnosed with Syndrome I accompanied by Syndrome III. In the early puberty group, Syndrome I was also the main syndrome (98.9% ). The degrees of most symptoms were mild to moderate. Reddened tongue was the most common tongue manifestation (6.86% prevalence) in the IPP group. The most common pulse manifestations were slippery pulse, thread pulse, and taut pulse. The Asymptotic Normalization Coefficient (ANC) method was used to quantify the TCM syndromes in 67 cases. Diagnostic accuracy rate reached 9%, comparable to expert diagnosis. CONCLUSION: We find that there are types of syndrome in the IPP group and in the early puberty group. Syndrome I (hyperactivity of fire due to Yin deficiency) is the main syndrome in the two groups. ANC may be an appropriate for quantification model of TCM syndromes. JTCM. All rights reserved. Key words: Puberty, precocious; Medicine, Chinese traditional; Symptom complex; Fuzzy mathematics; Cluster analysis; Asymptotic normalization coefficient INTRODUCTION Precocious puberty (PP) is the appearance of signs of puberty at an early age. Changes in mood are often observed and can be disruptive. Many families consider medium to long-term psychosocial effects of precocious puberty to be more important than the growth in height due to early epiphyseal closure, and early devel- JTCM www. journaltcm. com 6 October 5, Volume Issue 5

opment in girls may lead to unwanted sexual encounters, which may be more difficult to manage in children with learning difficulties. The incidence of precocious puberty has been increasing rapidly in recent years. In China, Traditional Chinese Medicine (TCM) is widely used. However, a standard TCM syndrome differentiation of precocious puberty has not yet to be established, which has hindered the development of TCM to some extent. The objective of this study is to investigate the distribution and characteristics of TCM syndrome in children with idiopathic precocious puberty (IPP) and early puberty through clinic data mining to establish a standard quantificational TCM model through fuzzy mathematics. MATERIALS AND METHODS General information One hundred and sixty-seven girls from the outpatient department of the Children's Hospital of Fudan University were divided into groups according to age. Patients under age 8 were classified into the group with idiopathic precocious puberty (n=5, 6.79±.); patients above 8 years old and under 9 years old were classified into the group with early puberty (n=6, 8.±.). This study was approval by Pediatric Research Ethics Board of Clinical Pharmacology Base, Fudan University in 9. All the patients in both groups voluntarily joined this study with informed consents. The criteria of cases were summarized as follows: Case-inclusion criteria included: 5 (a) IPP (the appearance of secondary sexual characteristics in girls below 8 years of age or the presence of menarche in girls under age 9; (b) no administration of sex hormones or Gn- RHa; (c) no previous consumption of Chinese herbal mixtures for medication. Case-exclusion criteria included: (a) organic lesions of the central nervous system caused by the central precocious puberty (CPP); (b) gonadotropin-independent precocious puberty, such as the McCune-Albright syndrome, testotoxicosis, primary hypothyroidism, gonadal tumors, and use of oral contraceptive; (c) heterosexual precocious puberty. The case was excluded if any of the above was diagnosed. Establishment of a TCM panel for precocious puberty A panel of TCM experts, who engaged in the clinical treatment of precocious puberty for more than years, was established to determine TCM syndrome differentiation in the children in accordance with the diagnostic criteria mentioned above through case discussions. Formulate questionnaires for classifying and quantifying 7 TCM symptoms The contents of this clinical epidemiological survey were to examine symptoms, signs, tongue, and pulse in children. According to the advice of many TCM experts, 7 symptoms were selected as characteristic symptoms, which were considered in diagnosis. Each symptom had levels (- point scale) that correspond with the patient's condition, with indicating symptom-free and indicating the most serious level (Table ). Precocious puberty was clinically divided into three syndromes, including hyperactivity of fire due to Yin deficiency (Syndrome I), depressed liver Qi that transform into fire (Syndrome II), and end retention of damp heat (Syndrome III). These conclusions were drawn by synthetizing the existing literature., There may also be intersections in the same coordinate domain, known as accompanying syndromes (Table ). Symptoms in each type of syndrome It was pointed out that disease differentiation and TCM syndrome differentiation were necessary to be considered. The relationship between symptoms in each type of syndrome was summarized as follows, which was referred to the Standard of Traditional Chinese Medicine Syndromes Diagnosis and Efficacy published in 995. 6 (a) The major symptoms of Syndrome II were dark urine/scanty dark urine, constipation, restlessness and irritability, depression, belching and sighing, stuffiness in chest and hypochondriac distension, breast swelling, insomnia, excessive dreams, tinnitus, dizziness/vertigo, heaviness in head, red eyes, distended eyes/pain in the eyes, dry eyes, bitter taste in mouth, halitosis, vellus around mouth, flushing, acne, reddened tongue tip, hypertrophied tongue and taut pulse. (b) The major symptoms of Syndrome III were spontaneous perspiration, recurrent fever, irregular watery stools, excessive dreams, tinnitus, sticky mouth, vellus around mouth, acne, sores and ulcers on skin, heavy feeling on the head and body, fatigue, lethargy, deficiency in Qi and no desire to speak, preference for greasy and sweet foods, abdominal distention after eating, morbid leucorrhea, thick tongue fur and slippery pulse. Statistical analysis A database was established and the results were analyzed using statistical software SPSS. (SPSS corporation, Chicago, IL, USA), the relationship between types of syndrome and 7 symptoms were summated. Meanwhile, the frequency distribution of each symptom or sign was aggregated. The frequency of the factor=the sum of frequencies for each factor/total %. Create fuzzy relationship matrix There were four levels to describe the severity of syndromes corresponding with the patient's condition, with indicating symptom-free and indicating the most serious level. The relationship between TCM symptoms and their types were obtained after combining the knowledge of TCM experts and the data of mean weighted processing from the cases confirmed by experts (Table ). Meanwhile a list of patients' diagno- JTCM www. journaltcm. com 6 October 5, Volume Issue 5

Table Questionnaires for classifying and quantifying TCM symptoms Hierarchical Item 5 6 7 8 9 5 6 7 8 9 5 6 7 8 9 5 6 7 8 9 5 6 7 Syndromes Dysphoria with feverish senation in chest, palms and soles' Fidal fever Hiding fever Fear of heat Night sweat Deep yellow urine/scanty dark urine Constipation Irregular sloppy stool Fidgetiness and irritability Depressiveness Belching and sighing Chest stuffiness and hypochondriac distension Breast swelling pain Insomnia Dreaminess Tinnitus Dizziness/vertigo Heaviness in head Red eyes Distending eyes/eye pain Dry eyes Bitter taste Halitosis Dryness of mouth Sticky mouth Red lips Vellus around mouth Flushing Reddening of the cheeks Acne Sore and ulcer on the skin Heavy feeling on the head and body Fatigue Spiritlessness Deficiency of Qi and no desire to speak Polyphagia (eat much) Fancy for greasy and sweet foods Abdominal distention after eating Morbid leukorrhea Reddened tongue Reddened tongue tip Hypertrophied tongue Thick tongue fur Greasy tongue fur Thready pulse Taut pulse Slippery pulse Symptom-free none Mild somtimes Moderate often Severe always JTCM www. journaltcm. com 6 October 5, Volume Issue 5

sis was set up (Table ), which was a case matrix. After divided by, the value of the patient diagnosis information table and the experts' knowledge table were limited to (, ) interval and then fuzzified. The fuzzy relationship matrix R about syndrome classification and 7 symptoms were obtained. So was the fuzzy relationship matrix X of patients and syndromes. Assume Z was a TCM syndrome, Y was syndrome classification of TCM, and R was fuzzy relationship matrix of X and Y, so Z ={z i i =,,...,m},y ={y j j =,,...,n} was established. ér r r n ù r R = ê r r n ú ê ú ër m r m r mn û Where the element value r ij = μ R (z i,y j ) was in the (, ) interval, (m=7, n=). So the matrix R was shown as Table. The matrix X was got in the same way. Algorithm of syndrome classification The diagnosis system could make out which syndrome class the case was. The diagnosis system gave out the results with different credibility value for each kind of syndrome classes. Used the and-product fuzzy synthesis, then the degree of membership that a patient l belonging to certain syndrome classification was showed: Table Relationship of 7 TCM symptoms and types of syndrome Types of syndrome ér r r ù êr x l x l... x lm r r ú ê ú =[b l,b l,b l] ê ú ër m r m r m û B l = X l R = [ ] r ij m b lj =( x li ),j =,,.i =,,,7 m i = r ij i = where w ij = r ij 7 i = r ij, i =,,,7; j =,, was the weight of each syndrome to certain syndrome classification. The degree of membership with the element value in matrix B determined the syndrome classification. The corresponding syndrome of max(b l,b l,b l ) was the first diagnosis result, and then the second and third diagnosis result in sequence. A percent form was used to express credibility of a patient belonging to a certain syndrome. b i % b i i = For example, if the fuzzy vector of a patient was B= (.7,.8,.), then the diagnosis result was: b > b > b. It meant that the first diagnosis was "Hyperactivity of fire due to Yin deficiency" with the % credibility degree; the second diagnosis was "dampness-heat stagnating in spleen" with the % degree; Symptom 7 Types of syndrome..6.7.6 Types of syndrome...5 Types of syndrome.76.8 Table Symptoms appear in children ID Name Symptom 7 case case case 67 case 67 Table Fuzzy relation matrix of TCM syndrome differentiation R Types of syndrome Types of syndrome Types of syndrome Syndrome.67.. Syndrome... Syndrome.567.. Syndrome 7..7. JTCM www. journaltcm. com 6 October 5, Volume Issue 5

the third diagnosis was "syndrome of stagnation of liver Qi inducing fire" with the 6% degree. To distinguish accompanied symptom in the algorithm, a threshold value ε was set up. If the difference of two symptoms b i - b j ε,the symptom i was the accompanied symptom of symptom j. Thus, the Asymptotic Normalization Coefficient (ANC) method of operation on fuzzy sets B was selected according to the diagnosis of the experts. RESULTS Frequency distributions of the 7 syndromes The syndromes of groups with IPP and early puberty were gathered from high frequency to low frequency separately, as shown in Tables and. The results showed that "discomfort in heat," "preference for greasy and sweet foods," "dysphoria with feverish sensations in the chest, palms, and soles," "dark urine/ scanty dark urine," "night sweats," "halitosis," "spontaneous sweating," and "restlessness and irritability" occurred at a higher frequency in the IPP group, with the frequency of "discomfort of heat" being 8.95% (Table 5). In contrast, the symptoms that had a higher frequency in the group with early puberty were: "halitosis," "dysphoria with feverish sensations in the chest, palms, and soles," "discomfort in heat," "restlessness and irritability," "breast swelling," "dark urine/scanty dark urine," "night sweats," "spontaneous sweating," "constipation, "and"pharyngitis" (Table 6). Thus, the symptoms present in both groups were "discomfort in heat," "preference for greasy and sweet foods," "dysphoria with feverish sensations in chest, palms, and soles," "dark urine/scanty dark urine," "night sweats," "halitosis," "spontaneous sweating," "restlessness and irritability," and "breast swelling." Table 5 Common symptoms of patients with IPP Characteristic symptom Discomfort in heat Preference for greasy and sweet foods Dysphoria with feverish senations in chest, palms, and soles Dark urine/scanty dark urine Night sweats Halitosis Spontaneous sweating Restlessness and irritability Breast swelling Polyphagia (overeating) Eye secretions Excessive dreams Lin YY et al. / Clinical Study Frequency (case) 85 8 7 6 6 6 6 58 58 5 5 8 The degrees of most symptoms were mild to moderate. Nevertheless, they occurred at different frequencies. Characteristics of tongue and pulse A reddened tongue was the most common symptom seen in the tongue (6.86% prevalence) in the group with IPP. More than half of the patients in the group (59.5%) had mild to moderate thick tongue fur. The most common pulse manifestations were slippery pulse, thready pulse, and taut pulse. The distribution of the group with early normal puberty was similar to that of the group with IPP (Table 7). Expert diagnosis of syndrome differentiations and combinations In the group with IPP, Syndrome I was the most common principal syndrome, occurring in all 5 girls. None of the girls had Syndromes II or III as principal syndromes. Forty-six girls (.8% ) had Syndrome I accompanied by Syndrome II and (.8% ) had Syndrome I accompanied by Syndrome III. Thirty cases (8.57% ) had all three syndromes, with Syndrome II as the main accompanying symptom, while cases (.8% ) had a combination of all three syndromes with Syndrome III as the main accompanying symptom. Similar to the IPP group, the group with early puberty also had Syndrome I as the main syndrome (98.9% of the 6 girls); patient was diagnosed with Syndrome III. The rest had a syndrome distribution similar to the IPP group; being diagnosed with only one syndrome was rare. Diagnosis by operation the fuzzy sets B The degree of membership, i.e., the element value in matrix B, determines the syndrome classification. The sequence in which the syndromes are entered into matrix B corresponds to the diagnosis. To confirm that a patient has a certain syndrome, a percentage was calcu- Percentage (%) 8.95 79.5 69.5 6.95 6 6 57. 55. 55. 5.8 8.57 5.7 Notes: only symptoms with frequencies above 5% are shown. IPP: idiopathic precocious puberty. Mild 6 56 5 7 5 Hierarchy (case) Moderate 5 6 7 5 5 8 7 8 Severe 6 JTCM www. journaltcm. com 6 October 5, Volume Issue 5

Table 6 Common symptoms of patients experiencing early puberty Characteristic symptom Frequency (case) Percentage (%) Mild Hierarchical (case) Moderate Severe Halitosis 67.7 5 Dysphoria with feverish senations in chest, palms, and oles 6.5 9 9 Discomfort in heat Restlessness and irritability Breast swelling Dark urine/scanty dark urine Night sweats Spontaneous sweating Constipation Pharyngitis 7 7 6 9 8 Note: only symptoms with frequencies above 5% are shown. Table 7 Characteristics of tongue and pulse IPP Tongue and pulse Characteristic symptom Frequency (case) 6.5 59.68 59.68 58.6 5.8 5.6 6.77 5.6 Percentage (%) 6 5 7 5 7 Early puberty Frequency (case) Percentage (%) Tongue Pulse Notes: only the top symptoms were presented. IPP: idiopathic precocious puberty. lated in the following manner. After using the ANC method to quantify the syndromes in 67 cases, we found that the diagnostic accuracy rate for the first diagnosis reached 96%, corresponding with the expert diagnosis. DISCUSSION Reddened tongue Reddened tongue tip Thick tongue fur Slippery pulse Thready pulse Taut pulse Precocious puberty as a specific name of disease has not been found in ancient TCM books or codes. In early 8s last century, after Wenhua Gu, the former chief pediatrician of division of Traditional Chinese Medicine at Children's Hospital of Fudan University, had investigated dozens of patients with IPP for TCM symptoms and signs, and found that many patients had clinical manifestations of unbalanced Yin and Yang in kidney and liver. So he first proposed the theory that the pathogenesis of IPP is deficiency of liver-yin and kidney-yin induced hyperfunction of ministerial fire. 7 Since then the application of Chinese herbs and preparations on IPP children has been gradually accepted nationwide. Although TCM is widely used in medical institutions, the lack of standardization for TCM syndrome differentiation through the investigation of the distribution and characteristics of TCM syndromes in children with IPP and early puberty limits the use of 66 57 6 5 TCM. It is demonstrated that using TCM has a unique advantage of treating precocious puberty, especially mild to moderate. Compared to GnRH analog injection, TCM treatment has fewer side effects and lower price. 8 The TCM makes a diagnosis and gives treatment according to syndrome differentiation. It is systemic and integrated, and gives multi-level and multi-angle deliberateness to the patient's condition. However the expression of the TCM concepts often uses a large number of fuzzy concepts, so the values representing the degree of the syndrome vary from person to person relying on the doctor's subjective judgment. As a result, it is lack of unified standards. It has become a main bottleneck that blocks the advanced development and modernization of the TCM. Therefore, some suggestions and solutions about the combination of TCM and fuzzy mathematics were proposed in the early 99s as well as using numerical computation and logical judgment synthetically. The application includes fuzzy diagnosis, fuzzy synthetic decision, threshold logic, pattern matching and so on. A small amount is used in TCM diagnosis model. 9- Then it appears other methods like approximate reasoning, uncertainty reasoning and classification inference. With the advanced research on the TCM theo- JTCM www. journaltcm. com 65 October 5, Volume Issue 5 6.86 5.9 59.5 9.5 8.57 9.5 5 8 56.5 5.6 5.8 7.97 5.8.6

ry, artificial intelligence technology is more frequently used. At that time, fuzzy technology application in TCM was still in a preliminary exploration, and there were not a lot of cases of application. Ding et al defined the conception of TCM knowledge and put it into knowledge base as well as put the inner link of the knowledge into rules based on the database technology and TCM expert system. The system can simulate the thoughts of Chinese medicine experts to a great degree for the use of reliability of reasoning. Since the late 99s, medical experts system began to combine the network technology, data mining technology, synergic and distributed system. It tends to a consulting system of large illness domain database.,-6 However, all kinds of intelligent diagnosis system has the same basic problem that is how to express the qualitative TCM knowledge in a quantitative way. As is well-known, we define different diseases according to the different degree of the syndrome in TCM. Patients with the same syndrome may be in a different degree, for someone may be heavy while others e mild. It will play an important role in keeping balance between groups of disease and judging the curative effects. This concept introduces fuzzy set theory which has a good advantage in expressing the qualitative TCM knowledge quantificationally and simulating the human reasoning process. Firstly, we revealed the multi-level relations between complex objects using fuzzy information granulation. Then, we discussed the extraction of the expert knowledge in TCM and developed fuzzy inference decisionbased on the acquired knowledge in quantitative case data. Finally, we used the ANC method to operate the fuzzy sets. As it was shown above, the model could be used to accurately screen for children with sexual precocity disease in this study with the accuracy rate of symptom classification reaching 96% in the 67 cases. Thus, ANC appears to be an appropriate method to operate a quantification model of TCM syndrome. However, the ANC algorithm requires improvement and more validation in future research. The function of the quantification model of TCM syndrome can be also expanded. Our data indicated that the characteristic symptoms of Syndrome I are: dysphoria with feverish sensations in the chest, palms and soles, night sweats, reddening of the cheeks, greasy tongue fur, and thready pulse in children. The symptoms that characterize Syndrome II are depression, belching and sighing, stuffiness in the chest and hypochondriac distension, breast swelling, bitter taste in mouth, hypertrophied tongue, and taut pulse. Abdominal distention after eating, thick tongue fur, and slippery pulse are characteristic symptoms of Syndrome III. However, the symptoms of the tongue and pulse that occur in the highest frequency in the distribution may not be consistent with the type of syndrome. Characteristic symptoms of the tongue and pulse, which one should be considered in the diagnosis, depend on the status of the disease as well as the experience of the expert. Therefore, it seems that characteristic syndromes may be more important than symptoms of the tongue and pulse. REFERENCES Loomba-Albrecht LA, Styne DM. The physiology of puberty and its disorders. Pediatr Ann ; (): E-E9. Kim EY, Lee MI. Psychosocial aspects in girls with idiopathic precocious puberty. Psychiatry Investig Mar; 9 (): 5-8. Sørensen K, Mouritsen A, Aksglaede L, et al. Recent secular trends in pubertal timing: implications for evaluation and diagnosis of precocious puberty. Horm Res Paediatr ; 77(): 7-5. Huang R, Wang YH, Yu J. Biliometric analysis of IPP treated with Chinese herbs in the past thirty years. Zhong Hua Zhong Yi Yao Za Zhi ; (): -5. 5 Chinese society of pediatric endocrinology and metabolism. Guideline for the diagnosis and treatment of central (true) precocious puberty. Zhong Hua Er Ke Za Zhi 7; 5(6): 6-7. 6 State Administration of Traditional Chinese Medicine. Criteria of diagnosis and therapeutic effect of Traditional Chinese Medicine (ZY/T.-9). Nanjing: Nanjing University Press, 995:. 7 Shi YM, Cao LH. Nourishing "Yin"-removing "Fire" Chinese medicinal herbs for girls with idiopathic precocious puberty. Liaoning Zhong Yi Za Zhi 98; ():. 8 Wang XF. Pediatrics of integrated Traditional and Western Medicine. Beijing: China Press of Traditional Chinese Medicine, : 7. 9 Li F, Cai LJ, Tian YZ. Medical diagnosis system based on vague sets. Hua Zhong Ke Ji Da Xue (Nature Science) ; (): 7-9. Li A. Fuzzy mathematics and typing diagnosis of cervical spondlosis. Zhong Guo Zhong Yi Gu Shang Ke Za Zhi 996; 8(): 6-7. Ding M, Li YP. Advancement in research of intellectualized TCM diagnosis. Jiangsu Zhong Yi Za Zhi 9; (8): 77-79. Zhou CG, Zhang ZF. Progress and prospects of research on information processing techniques for intelligent diagnosis of Traditional Chinese Medicine. Zhong Xi Yi Jie He Xue Bao 6; (6): 56-566. Yu JW, Ma LZ, Yang HY. Present research situation and prospect for intelligent diagnosis of Traditional Chinese Medicine. Liaoning Zhong Yi Za Zhi ; 7(): 5-5. Hao WS, Zhu XS, Wang XR, et al. Biochemical index estimation from bioelectric signal information-the ANFIS method. Zhong Guo Lin Chuang Yi Sheng Za Zhi 8; (): 9-98. 5 Yang YP, Hu JJ. Application of fuzzy cognitive-map in collaborative medical diagnosis system. Ji Suan Ji Gong Cheng Yu Ying Yong 6; (7): 8-. 6 Yang LC. Talk about Chiniese medicine experts system from the Internet. Hunan Zhong Yi Za Zhi 999; 9(): 6-6. JTCM www. journaltcm. com 66 October 5, Volume Issue 5